# 10/17/24 Meeting Notes #9

# This Week's Progress 這週進度
## Classification Model
- `Github` [HERE](https://github.com/Vyo918/Undergraduate_Project)
-- The code is in main.py of "Classification_Model_1" [HERE](https://github.com/Vyo918/Undergraduate_Project/tree/main/Classification_Model_1/main.py) and "Classification_Model_2" [HERE](https://github.com/Vyo918/Undergraduate_Project/tree/main/Classification_Model_2/main.py)
- Current paper used for Classification model #1
`目前用於分類模型 #1 的論文`
-- **Classification of Garments from Fashion MNIST Dataset Using CNN LeNet-5 Architecture** [HERE](https://ieeexplore.ieee.org/document/9047776)
- Have tested the code of Classification Model #1 using Fashion MNIST dataset.
`已使用 Fashion MNIST 數據集測試分類模型 #1 的程式碼`
-- The test accuracy is 90.68%. `測試準確率為 90.68%`

## Generative Model
- Github for the paper. `論文的 Github` [HERE](https://github.com/HumanAIGC/Cloth2Tex)
-- Installed `已安裝`: Pytorch3D, torch_geometric, kaolin
-- Current Progress `目前進度`: installing psbody

## Datasets
- Github [HERE](https://github.com/BS3L/EMPS-datasets)
-- This github will have all the datasets for each model.
`這個 Github 將包含所有模型的數據集`
-- All are currently incomplete, still in the process of separating datasets and uploading datasets.
`目前都尚未完成，正在分離和上傳數據集中`

- Uploaded datasets for classification model#1, in train -dress, pants, shorts, flats, heels, shoes, sneakers, t-shirt; in test -pants.
`已上傳的分類模型 #1 數據集，訓練集中 - dress, pants, shorts, flats, heels, shoes, sneakers, t-shirt；測試集中 - pants`
-- Other datasets for classifcation model#1 train still uploading. For test, still separating and finding more datasets.
`其他分類模型 #1 的訓練集數據仍在上傳中。測試集部分還在分離和尋找更多數據集中`

## Dataset for classification model #2
- Currently separating datasets.
`目前正在分離數據集中`
- Dataset from **Fashion IQ: A New Dataset Towards
Retrieving Images by Natural Language Feedback (CVPR 2021)**
-- `Paper` [HERE](https://openaccess.thecvf.com//content/CVPR2021/papers/Wu_Fashion_IQ_A_New_Dataset_Towards_Retrieving_Images_by_Natural_CVPR_2021_paper.pdf)
-- Dress: 19088 (18134 成功下載successfully downloaded)
[Individual dress piece 268]
– Shirt: 31729 (30837 成功下載successfully downloaded)
– Tops & tees: 26869 (26274 成功下載successfully downloaded)
-- Still need to filter through the dataset ourselves. 仍需我們自己過濾數據集。

- Dataset from **Fahion144k**
-- Dataset containing 144,149 user posts.
`包含 144,149 用戶貼文的數據集`
-- `Paper` [HERE](https://paperswithcode.com/paper/neuroaesthetics-in-fashion-modeling-the-1)
-- Website [HERE](https://esslab.jp/~ess/en/data/fashion144k/)
-- Still need to label the dataset ourselves.
`仍需自行標註數據集`

## To Do For Next Week
- Start using our own datasets to train classifcation model #1.
- Test classification model #2 using paper's dataset and check its accuracy.
- Finish setting up the environment for generative model.
- Determine if we will need to use the lab's server. If yes, we need to learn how to use Docker.

---
Previous: [10/10 Meeting Notes #8](https://hackmd.io/@emps-113up/meeting8)
Next: [10/24 Meeting Notes #10](https://hackmd.io/@emps-113up/meeting10)
Full Content List [here](https://hackmd.io/@emps-113up/full-list)